Deficits in Movement Planning and Intrinsic
Coordinate Control in Ideomotor Apraxia
Steven A. Jax1,2
, Laurel J. Buxbaum1,3
, and Adrienne D. Moll1
Abstract
& Two central issues in the field of motor control are the
coordinate frame in which movements are controlled and
the distinction between movement planning and online correction.
In this study we used these issues to frame several
hypotheses about the deficits underlying ideomotor apraxia
(IMA). In particular, we examined whether ideomotor apraxics
exhibited (1) deficits in movement control in intrinsic
(body relative) coordinates with better control in extrinsic
(workspace relative) coordinates, (2) deficits in movement
planning that are compensated for by an overreliance on
online correction, or (3) both deficits. Patients with IMA
and two comparison groups performed movement tasks that
relied preferentially on either intrinsic or extrinsic coordinate
control when online correction was either possible
or impossible. Participants performed posture imitation and
grasp imitation movements to body- and object-relative end
positions in the presence or absence of visual feedback. Consistent
with the intrinsic coordinate control hypothesis, patients
with IMA showed a significantly greater disparity than
the other two groups between movements made to bodyrelative
and object-relative targets as well as between imitation
of meaningless postures and grasping. Consistent with
the correction overreliance hypothesis, the IMA group was
more disrupted than the other groups by the removal of
vision. Thus, IMA patients exhibit behavioral patterns consistent
with both deficient intrinsic coordinate control and overreliance
upon visual feedback. Finally, lesion analysis suggests
that damage to the left inferior parietal lobe (Brodmann’s
areas 39 and 40) may play a key role in both behavioral
deficits. &
INTRODUCTION
One of the central issues in the field of motor control
is the coordinate frame in which movements are controlled
(for a recent review, see Vindras, Desmurget, &
Viviani, 2005). Many theories have proposed that movements
are controlled in workspace-specified extrinsic
coordinates, such as the spatial vector describing a
desired movement’s direction and amplitude (Vindras
& Viviani, 1998; Gordon, Ghilardi, & Ghez, 1994). These
accounts are contrasted with those that propose control
in body-specified intrinsic coordinates, such as the
shoulder and elbow angles during reaching (Rosenbaum,
Meulenbroek, Vaughan, & Jansen, 2001). Extensive evidence
for both types of control has led to a third group
of accounts proposing that control is an interactive
process that uses both extrinsic and intrinsic coordinate
frames (Van Thiel, Meulenbroek, & Hulstijn, 1998;
Kawato, 1996). Although dual coordinate frame accounts
can accommodate evidence for control in either coordinate
frame, they often lack a clear explanation for the
utility of such a strategy. One possibility is that dual coordinate
control may allow the motor system to flexibly
utilize the most appropriate coordinate frame for a
particular task. For example, Ghafouri, Archambault,
Adamovich, and Feldman (2002) suggested that movements
to stationary targets are controlled in extrinsic coordinates,
whereas movement to targets that move along
with the body are controlled in intrinsic coordinates.
Additional support for the use of task-specific coordinate
frames comes from the pattern of movement
deficits exhibited by patients with ideomotor apraxia
(IMA). These patients are commonly impaired in pantomiming
and imitating object-related actions, such as
hammering or sawing, when the object is not present
(Buxbaum, Giovannetti, & Libom, 2000; Poizner et al.,
1998) but perform more accurately with the object
in hand (Goldenberg, Hentze, & Hermsdo¨rfer, 2004;
Heilman & Gonzalez Rothi, 1993). Previous explanations
of this difference have focused on the additional tactile
feedback available when objects are held (Poizner et al.,
1998; but see Goldenberg et al., 2004, for conflicting
evidence). An alternative, although not mutually exclusive,
explanation is that pantomime and imitation
without an object are strongly weighted toward intrinsic
control (e.g., specifying joint rotations at the shoulder
and elbow to make a hammering motion), whereas
object-related actions may be more strongly weighted
toward extrinsic control (e.g., specifying the trajectory
1
Moss Rehabilitation Research Institute, Philadelphia, 2
University
of Pennsylvania Medical School, 3
Thomas Jefferson University,
Philadelphia
D 2006 Massachusetts Institute of Technology Journal of Cognitive Neuroscience 18:12, pp. 2063–2076
of the hammer through space). On this assumption, the
pattern of performance in patients with IMA is consistent
with deficits in intrinsic coordinate control in the
face of relatively intact control in extrinsic coordinates
(Buxbaum, 2001; Buxbaum et al., 2000).
Two other findings in IMA are consistent with this intrinsic
coordinate control hypothesis. First, patients with
IMA commonly exhibit difficulties when imitating meaningless
body-related movements, such as placing a horizontally
oriented hand below the chin (Goldenberg &
Hagmann, 1997; De Renzi, 1985). Like imitation and
pantomime without an object in hand, performing meaningless
movements requires the specification of a coordinated
pattern of joint rotations, an intrinsic coordinate
frame control process. Second, patients with IMA exhibit
relatively intact grasping of objects (Buxbaum, JohnsonFrey,
& Bartlett-Williams, 2005; Hermsdorfer, Ulrich,
Marquardt, Goldenberg, & Mai, 1999). In the present account,
this task is primarily performed by specifying a
spatial vector (an extrinsic coordinate property) between
the location of the hand and the location of the object
to be grasped. Computational models using this method
of control can simulate reaching well (Bullock, Grossberg,
& Guenther, 1993; Bullock & Grossberg, 1988).
Although the distinction between intrinsic and extrinsic
coordinate frames has been central in the study of
movement control, the process of control itself is commonly
subdivided into planning and online correction
components (Glover, 2004; Elliot, Helson, & Chua, 2001;
Woodworth, 1899). Planning is the preparation of a
movement plan before movement initiation, and online
correction is the adjustment of the movement plan
during movement execution. We have recently proposed
that IMA may be attributable in part to deficits
in planning actions with relatively intact online correction
(Buxbaum, Johnson-Frey, et al., 2005). To support
this claim, we compared performance on a grasping
imagery task, which entailed movement planning, but
not correction, to actual grasping, which entailed both
planning and correction. The results showed that patients
with IMA exhibited abnormal imagined grasping
with relatively intact actual grasping. Similar motor
imagery deficits have been reported by Sirigu et al.
(1996). Additional evidence that patients with IMA have
deficits in movement planning but not online correction
is that reaching performance deteriorates in IMA patients
much more than in controls when online correction
is not possible, such as when vision of the arm
or object is not available (Laimgruber, Goldenberg, &
Hermsdorfer, 2005; Haaland, Harrington, & Knight, 1999;
although see Ietswaart, Carey, Della Sala, & Dijkhuizen,
2001, for conflicting results). Thus, not only do IMA
patients have relatively intact online correction abilities,
they also seem to be overly reliant on them.
The goal of the current study was to jointly test the
intrinsic coordinate control and correction overreliance
hypotheses by examining how patients with IMA performed
on tasks that rely preferentially on either intrinsic
or extrinsic coordinate control when online
correction was either possible or impossible. Three
groups of participants were assessed: patients who had
suffered left cerebral vascular accidents (LCVAs) and
exhibited IMA (hereafter, the IMA group), patients
who had LCVAs without IMA (hereafter, the LCVA
group), and neurologically intact age-matched controls
(hereafter, the control group). All participants were
tested on grasping and meaningless posture imitation
tasks. Meaningless, and not meaningful, posture imitation
was used because it allowed us to examine the
movement production abilities of patients with IMA
without the possible influence of damage to stored gesture
representations (Goldenberg & Hagmann, 1997).
For both tasks we manipulated how movement end
positions were defined. Half of the movements were
made to end positions defined by an object in the environment
(e.g., grasping a bottle; producing a stop hand
posture above a pair of scissors), and the other half were
made to end positions defined by the participant’s body
(e.g., grasping the earlobe, producing a stop hand posture
to the left of the left ear). All four movement types
(object-relative posture, object-relative grasp, bodyrelative
posture, body-relative grasp) were tested with
and without visual feedback, thus manipulating the possibility
of online correction.
Given that IMA patients exhibit particular deficits in
the production of accurate hand postures relative to other
movement components (Buxbaum, Kyle, Grossman, &
Coslett, in press; Buxbaum, Sirigu, Schwartz, & Klatzky,
2003; Sirigu et al., 1996), movement production was
measured by using four hand-related subcomponents.
Two of the components (hand configuration and wrist
angle) were putatively specified in an intrinsic coordinate
frame, whereas the other two (orientation of the
hand in space and location of the hand) were putatively
specified in an extrinsic coordinate frame. Finally, to
ensure that deficits on these tasks could be ascribed to
movement control processes and not more elementary
deficits in perception or working memory, we included
a movement perceptual matching task that had no
movement production component.
We derived several predictions from each hypothesis.
The intrinsic coordinate control hypothesis predicts that
the IMA group (as compared to the control and LCVA
groups) would be (1) less accurate on posture imitation
than grasp imitation due to the relative reliance on
intrinsic control during posture imitation, (2) less accurate
on body-directed than object-directed movements
due to the relative reliance on extrinsic control for
object-directed movements,1
(3) more impaired in movement
components specified in intrinsic than extrinsic
coordinates, and (4) unaffected by the removal of vision
during movement production.
The correction overreliance hypothesis, in contrast,
predicts that the IMA group (as compared to the control
2064 Journal of Cognitive Neuroscience Volume 18, Number 12
and LCVA groups) would be (1) equally good at grasp
and posture imitation, (2) equally good at movements
made to body-relative and object-relative end positions,
(3) equally good at all movement components, and (4)
more negatively affected by the removal of vision.
Clearly, the two hypotheses described above are not
mutually exclusive, and thus a final possibility is that
both hypotheses are correct, which would predict effects
of all the factors examined.
Previous studies indicate that IMA most often results
from damage to the left inferior parietal lobule (IPL;
Brodmann’s areas [BAs] 39 and 40; Buxbaum, JohnsonFrey,
et al., 2005; Buxbaum, 2001; Haaland, Harrington,
& Knight, 2000; Heilman, Gonzalez Rothi, & Valenstein,
1982). However, it has also been reported to result
from damage to the left middle frontal gyrus (BAs 6,
8, 9, and 46; Haaland et al., 2000) as well as the superior
parietal lobes (BAs 5 and 7), which may be especially
important for action production but not action
recognition (Buxbaum et al., 2000; Rapcsak, Ochipa,
Anderson, & Poizner, 1995; Heilman, Gonzales Rothi,
Mack, Feinberg, & Watson, 1986). In addition to areas
that are commonly damaged in IMA, neuroimaging
studies have identified the importance of the inferior
prefrontal cortex (BAs 44 and 45) in action production
during imitation (Iacoboni, 2005; Johnson-Frey et al.,
2003). Guided by this work, we examined which, if
any, of these regions were significantly associated with
movement production deficits on the tasks used in the
present experiment.
METHODS
Participants
Eleven post-acute LCVA patients with IMA, four postacute
LCVA patients without IMA, and six neurologically
intact participants completed the experiment. All participants
were right-handed. IMA was diagnosed by averaging
scores on 10 trials of pantomime to sight of object
and 10 trials of meaningless posture imitation, all of
which were performed using the left arm. Each trial was
scored on four movement components (hand posture,
arm posture, amplitude, and timing), each of which
could be scored correct or incorrect (see Buxbaum
et al., 2000, for details of the scoring criteria). On these
tasks, all patients in the IMA group scored two standard
deviations or more below the mean of a group of 10 agematched
controls (see Buxbaum, Kyle, & Menon, 2005,
for details). Scores on these tests for both the IMA and
LCVA groups of the present study, along with other
participant information, are shown in Table 1. Using
single-factor analyses of variance (ANOVAs), we found
that the three groups did not differ in age ( p = .48) or
education ( p = .13), and the IMA and LCVA groups did
not differ in the time since stroke onset ( p = .94) or
lesion volume ( p = .20). All participants consented to
the study in accordance with the guidelines of Albert
Einstein Medical Center Institutional Review Board and
were paid for their participation.
Materials, Design, and Procedure
Movement Production Tasks
In the movement production tasks, participants imitated
movements made to static end positions using their left
arms. Ten different movements from each of the four
possible production conditions (object-relative posture,
object-relative grasp, body-relative posture, body-relative
grasp) were used. The postures used were similar to
those used by Goldenberg and Hagmann (1997), and
the objects used to define the end positions were
common objects (mug, pen, bottle, etc.). The movements
to be imitated were presented by using video
clips displayed on a computer screen located approximately
100 cm in front of the participant. Each video clip
was approximately 3.5 sec long and was presented twice
with a 1-sec delay between clips. Because the participants’
ability to reproduce the final position was of
primary interest in the study, each clip was created so
that the final position was maintained for 2.5 sec in all
videos.
For all four movement production conditions, there
were three different movement initiation conditions.
The first initiation condition permitted participants to
initiate movements as they watched the video clips
(immediate initiation). A second and third condition
were used to directly compare movement production
with and without possible online correction. In the
second condition, where online correction was not
possible, participants were blindfolded at the conclusion
of the second video, and responded when they heard a
tone that sounded 5 sec after the end of the second
video (delayed blindfolded initiation). The third condition,
which served as a control for the delay aspect of
the second condition required to position the blindfold,
was identical to the delayed blindfolded condition except
that subjects were not blindfolded (delayed initiation).
Each participant completed 10 trials in each of
the 12 conditions: 3 (initiation levels) Â 2 (movement
tasks) Â 2 (end positions). The initiation factor was
blocked and counterbalanced across participants; movement
task and end position factors were randomized
within blocks.
If participants made any observable movements before
the tone in either the delayed or delayed blindfolded
conditions, the experimenter reminded the participant
to wait for the tone. Trials in which this occurred (<7%
of trials) were rerun at the end of the block. To ensure
they understood the task, participants were given two
practice trials the first time they were tested in each of
the three initiation conditions, during which they were
given feedback about their performance. All practice
Jax, Buxbaum, and Moll 2065
trials used movements that were not used in the later
experimental trials.
Movement Matching Task
In the movement matching task, participants watched a
pair of short video clips, each showing an actor making a
movement. The two clips were presented sequentially
with 1 sec of blank screen between them. Immediately
after the presentation of the second clip, the screen was
again blanked and the experimenter asked the participant
if the movements in the clips were the same or
different. The participants’ task was to verbally respond
to this question. If patients had difficulties making verbal
responses, the experimenter had them point to a sheet
of paper with two boxes, one labeled ‘‘yes’’ (indicating
the clips were the same) and one labeled ‘‘no’’ (indicating
the clips were different). In the event that no
response was given within 5 sec, the experimenter
encouraged the participant to make a guess.
Table 1. Participant Information
Group Subject Sex Age
Education
(years)
Months
Poststroke
Pantomime to
Sight of Object (%)
Meaningless
Imitation (%)
Lesion Volume
(mm3
)
IMA I1 M 61 16 81 55.0 47.5 148,831
I2 F 70 12 47 72.5 65.0 8,501
I3 M 82 11 51 60.0 50.0 40,902
I4 M 53 9 19 87.5 75.0 32,695
I5 M 62 12 165 77.8 72.5 146,472
I6 F 59 12 22 80.0 82.5 a
I7 M 72 8 156 87.5 65.0 269,930
I8 M 61 9 32 87.5 65.0 101,058
I9 M 62 12 11 75.0 72.5 14,871
I10 F 51 16 76 82.1 65.0 95,940
I11 M 59 18 60 70.0 75.0 266,061
Mean 62.9 12.3 65.5 75.9 66.8 112,526
SD 8.8 3.2 52.1 11.0 10.6 96,082
LCVA L1 M 67 11 120 95.0 90.0 137,337
L2 M 54 10 51 95.0 95.0 7,927
L3 M 57 14 9 92.5 95.0 12,760
L4 F 47 13 91 97.5 95.0 11,790
Mean 56.3 12.0 67.7 95.0 93.8 42,453
SD 8.3 1.8 48.5 2.0 2.5 63,290
Control C1 F 48 16
C2 F 74 16
C3 M 74 13
C4 F 55 13
C5 F 58 16
C6 F 62 16
Mean 61.8 15.0 90.9b
94.3b
SD 10.5 1.6 5.5b
4.7b
a
Subject was excluded from lesion analysis, see Footnote 3.
b
Data reported for 10 previously tested age-matched controls used to identify apraxia in the current study (from Buxbaum et al., 2005).
2066 Journal of Cognitive Neuroscience Volume 18, Number 12
Forty pairs of matching task clips were used. One of
the clips within the pair was taken from the 40 clips used
in the movement production tasks described above. In
half of the second clips, the same actor produced the
same movement a second time (‘‘same’’ trials). In the
other half of the clips, the actor produced a slightly
different movement (‘‘different’’ trials). In the matching
trials of the grasp task, the movement in the ‘‘different’’
clip required the same final hand configuration but
grasped the target (the object in the object-relative
end positions or the body part in the body-relative
end positions) in a different location. For example, if
the first clip required a grasp of the center of a telephone
headset, the second clip required a grasp of the
end of the headset. In the matching trials of the posture
task, the movement in the ‘‘different’’ clip required the
same final hand location but a different hand posture.
For example, if the first clip required the production of
a stop hand posture above a telephone headset, the
second clip required a hand posture ending above the
phone but with the fingers together and pointing to
the right with the palm facing the participant’s body
(using the left hand).
Scoring Criteria for Movement Production Trials
A videotape of each participant’s performance was analyzed,
and the final hand posture on each movement
production trial was scored on four components: hand
configuration, wrist angle, hand orientation, and hand
location. Each of the four components could be scored
as correct (1) or incorrect (0). The hand configuration
component was scored as correct if the hand was
correctly configured or not flagrantly misconfigured
(e.g., fingers were slightly spread when the target hand
configuration required all fingers be pressed tightly
together). The wrist angle component was scored as
correct if the hand would have to be rotated less than a
total of 308 along any of the wrist’s axes to be correct.
The hand orientation component was scored as correct
Figure 1. Examples of performance by patients in the IMA group. (A) and (D) illustrate the target actions for the body-relative posture task
and object-relative grasp task, respectively. (B) Exemplifies the deficits of IMA patients on the body-relative posture task (the task which
putatively relied most strongly on intrinsic coordinate control), even when online correction was possible (black covering over face added
during figure preparation to mask the patient’s identity). (E) Exemplifies accurate performance on the object-relative grasp task when online
correction was possible (the task that putatively relied most strongly on extrinsic control). Accuracy was reduced without online correction
as compared to when online correction was possible for both the body-relative posture (C) and object-relative grasp tasks (F).
Jax, Buxbaum, and Moll 2067
if the orientation of the hand posture could be made
correct with a rotation of less than 308. The hand
location component was scored as correct if the center
of the hand would be in the correct end location with a
movement that was less than 30% of the length of the
target body part or object. Two different coders independently
scored the movements from three participants,
and the percent agreement between coders for
each of the four criteria was high (94% agreement or
higher for all four criteria).
Lesion Analysis
Lesions were segmented by an experienced research
team member under the supervision of a neurologist
(who was blind to the behavioral data) using the MRIcro
image analysis program (www.psychology.nottingham.
ac.uk/staff/cr1/mricro.html). Reliability in lesion segmentation
across team members was established by using
a method devised by Fiez, Damasio, and Grabowski
(2000). All team members achieved reliability comparable
to or better than that of Fiez et al. (2000) on three
measures: percent volume difference (7.99%), percent
discrepant voxels (4.18%), and percent nonoverlapping
voxels (31.5%). After the lesions were segmented, proportion
of damage to Brodmann’s areas was calculated
by using Brodmann maps included in the MRIcro program,
which were based on the templates developed by
Damasio and Damasio (1989). The proportion of damage
to larger regions of interest were computed as the
average damage to the region’s component Brodmann’s
areas (e.g., BAs 39 and 40 for the IPL), weighted by the
number of voxels in each area.
RESULTS
All effects of interest were assessed with mixed-factor
ANOVAs, where all factors were manipulated within
participants except the group factor (control, LCVA, or
IMA). When required, post hoc comparisons were performed
using Bonferroni-corrected t tests.
Movement Production with Online Correction
We first assessed the predictions relevant to movement
production with full vision (immediate initiation condition),
in which online correction was possible. Analyses
focused on the effects of Movement Task (grasp or
posture) and End Position (object relative or body
relative). Examples of performance by patients in the
IMA group in this condition are shown in the middle
column of Figure 1.
Mean Score
We first examined the mean scores (average of the four
component scores) for all movement production tasks
(Figure 2). There was a main effect of Group, F(2,18) =
14.00, p < .001, with mean scores being lower for the
IMA group than the control ( p < .001) and LCVA ( p =
.005) groups, which did not differ from one another
( p = .69). In addition to the main effect of Group,
there was a main effect of End Position, F(1,18) =
38.42, p < .001, and an interaction between Group and
End Position, F(2,18) = 5.26, p = .016. Although all three
groups had lower scores on body-relative end positions
than object-relative end positions ( p .042), the differFigure
2. Mean scores
(±1 SE) for all movement
production tasks. Separate
panels are used for the
immediate, delayed,
and delayed blindfolded
conditions. Within each panel,
data are plotted separately
for the grasp (G) and posture
(P) imitation movements
made to object-relative (OR)
and body-relative (BR) end
positions.
2068 Journal of Cognitive Neuroscience Volume 18, Number 12
ence between the end positions was larger in the IMA
group (.75 vs. .91) than the LCVA (.88 vs. .97, p = .048)
and control (.91 vs. .96, p = .018) groups, which did
not differ ( p = .43). Finally, there was an interaction
between Movement Task and End Position, F(1,18) =
11.35, p = .003, in which there was an effect of Movement
Task (posture < grasp) for object-relative end
positions ( p = .002), but not for body-relative end positions
( p = .40). All other remaining effects were not
reliable ( p ! .20).
Component Scores
Next, we examined how the individual components (hand
configuration, wrist angle, hand orientation, hand location)
contributed to the effects reported for the mean
scores. To do so, we repeated the mean score analysis
separately for each of the four component scores. For
conciseness, we will focus on the significant effects of
Group and the interactions between Group and other
factors in these ANOVAs, which are plotted in Figure 3.
Main effects of Group (Figure 3A) were observed for
all components ( p .02) except the hand configuration
component, which did not reach significance ( p = .13).
For the three components with main effects of Group,
the IMA group had lower scores than the LCVA
( p .037) and control groups ( p .011), which did
not differ ( p = .21).
Finally, although no interaction between Group and
Movement Task had been observed for the mean score
analysis reported earlier, this interaction was significant
for the wrist angle component, p = .017 (Figure 3B). An
effect of Movement Task was observed for the IMA,
p < .001, and control, p = .01, groups, with the
magnitude of the difference between posture and grasp
tasks being greater for the IMA group (.85 vs. .96) than
the control (.95 vs. .99) group ( p = .045).
Movement Production without Online Correction
Next, we assessed the predictions relevant to movement
production without visual feedback (delayed blindfolded
Figure 3. Mean scores (±1
SE) for the significant effects of
the component score analyses
of the immediate (A–B) and
delayed/delayed blindfolded
(C–F) conditions. (A, C) Main
effects of group for the hand
configuration (HC), wrist angle
(WR), hand orientation (OR),
and hand location (LOC).
(B, D) Interaction between
group and movement task for
the wrist angle component.
(E) Interaction between group
and initiation for the hand
configuration component.
(F) Interaction between group
and end position for the
hand orientation component.
Jax, Buxbaum, and Moll 2069
initiation condition), in which online correction was impossible.
Example movements made by IMA patients in
this condition are shown in the right column of Figure 1.
The delayed blindfolded initiation condition was compared
to a second condition (delayed initiation condition)
in which online correction was possible and, unlike
the immediate imitation condition, controlled for the
delay aspect required to position the blindfold.2
Analyses
focused on the effects of Initiation (delayed or delayed
blindfolded), Movement Task (grasp or posture), and
End Position (object relative or body relative).
Mean Score
As was observed in the immediate initiation condition,
there was a main effect of Group on mean scores,
F(2,18) = 22.86, p < .001, with the IMA group having
lower scores than the control ( p < .001) and LCVA
( p = .001) groups, which did not differ from one
another ( p = .56; see Figure 2).
In addition to the main effect of Group, there was a
main effect of Initiation, F(1,18) = 13.24, p = .002, and
an interaction between Group and Initiation, F(2,18) =
4.13, p = .03. Effects of Initiation were only observed
for the IMA group, with scores being lower in the
delayed blindfolded than the delayed initiation condition
(.78 vs. .83, p < .001).
Replicating the results in the immediate initiation
condition, there was a main effect of End Position,
F(1, 18) = 55.77, p < .001, and an interaction between
Group and End Position, F(2, 18) = 8.36, p = .003.
Although all three groups had lower mean scores for
body-relative than object-relative end positions, this difference
was larger in the IMA group (.74 vs. .87) than the
LCVA (.88 vs. .94, p = .04) and control groups (.90 vs.
.94, p = .009), which did not differ ( p = .52).
Finally, there was a main effect of Movement Task,
F(1, 18) = 14.01, p = .001, and an interaction between
Movement Task and End Position, F(1, 18) = 16.86,
p < .001, with effects of Movement Task being observed
for object-relative end positions, p < .001, but not for
body-relative end positions, p = .40. All other remaining
effects were not reliable, p ! .225.
Component Scores
As was done for the immediate initiation condition, we
repeated the mean score analysis separately for each of
the four component scores. Again, we will focus on the
significant effects of Group and the interactions between
Group and other factors in these ANOVAs, which are
shown in Figure 3.
Main effects of group were observed for all components
( p .016; Figure 3D). For each of the components,
the IMA group had lower scores than the LCVA
( p .045) and control ( p .041) groups, which did
not differ ( p = .65).
As in the analyses for the immediate condition, although
no interaction between Group and Movement
Task had been observed for the mean score, this interaction
was significant for the wrist angle component
( p = .02; Figure 3D). Effect of Movement Task were
only observed for the IMA group ( p < .001) and control
( p = .04) groups. However, the magnitude of difference
between Posture and Grasp was greater ( p = .005)
for the IMA group (.87 vs. .95) than the control (.96
vs. .99) group.
The interaction between Group and Initiation was
only significant for the hand configuration component
( p = .04; Figure 3E). Effects of Initiation were only observed
for the IMA group ( p = .001), with hand configuration
scores being lower in the delayed blindfolded
than the delayed initiation condition (.80 vs. .87).
The interaction between Group and End Position was
only significant for the hand orientation component
( p = .018; Figure 3F). Although hand orientation scores
were lower for body-relative than object-relative end
positions in all groups, the difference was larger for
the IMA group (.66 vs. .88) than the LCVA (.86 vs. .96,
p = .033) and control (.85 vs. .96, p = .042) groups,
which did not differ ( p = .86).
Movement Matching
Results of the movement matching task are shown in
Figure 4. Overall, all participants performed quite accuFigure
4. Mean proportion correct (±1 SE) for all movement
matching tasks. Data are plotted separately for the grasp (G) and
posture (P) video clips showing movements made to object-relative
(OR) and body-relative (BR) end positions.
2070 Journal of Cognitive Neuroscience Volume 18, Number 12
rately on this task. The results of a 3 (group: control,
LCVA, or IMA) Â 2 (Movement Task: grasp or posture) Â
2 (End Position: object-relative or body-relative) ANOVA
showed that the only reliable effect was a main effect of
End Position, F(1,18) = 4.43, p = .05, with accuracy on
matching body-relative movements being lower than
object-relative movements (.95 vs. .99). The main effect
of Group was not reliable, F(2,18) = 2.28, p = .131,
nor were any interactions between Group and the other
factors ( p = .21).
Lesion Analysis
Finally, we explored the neuroanatomic substrates of
patients’ movement production deficits in two ways.
First, we compared lesions in the IMA and LCVA groups
to assess whether IMA in the present group could be
ascribed to lesions similar to those previously reported
(Buxbaum, Johnson-Frey, et al., 2005; Buxbaum, 2001;
Haaland et al., 2000). Second, we compared lesion loci
of high and low performers on the experimental task
to explore the neuroanatomic structures mediating
successful performance. Given the reliable behavioral
differences between the IMA and LCVA groups, the
two analyses would not be predicted to be vastly different.
However, the second analysis allowed us to
examine which neural structures were most strongly
relied upon to successfully perform the imitation tasks
we used.
Comparison of IMA and LCVA Groups
The left side of Figure 5 displays a lesion subtraction
analysis comparing the IMA3
and LCVA groups. Consistent
with previous studies, IMA was associated with
damage to the IPL (BAs 39 and 40) and portions of the
middle frontal gyrus (BAs 6 and 46). In addition to these
previously reported areas, damage to both cortical and
subcortical areas of the superior temporal lobes (BAs 22,
41, and 42) was more common in the IMA group than
the LCVA group. The preceding lesion subtraction analyses
were confirmed statistically by comparing the average
proportion of damage in the IMA and LCVA groups
in the four a priori defined regions reviewed in the
Introduction: (1) the superior parietal lobule, BAs 5 and
7, (2) the middle frontal gyrus, BAs 6, 8, 9, and 46, (3)
the inferior prefrontal cortex, BAs 44 and 45, and (4) the
IPL, BAs 39 and 40. Using a nonparametric hypothesis
test,4
we found that only in the IPL was damage significantly
more extensive in the IMA group than the LCVA
group (.42 vs. .08, p = .021; p ! .21 for the other three
regions). In addition to the a priori defined regions,
damage to the superior temporal gyrus (BAs 22, 41, and
42) was greater in the IMA group than the LCVA group
(.44 vs. .10, p = .034).
Figure 5. Results of lesion subtraction analysis comparing the IMA and LCVA groups (left) as well as the groups with low and high movement
production scores (right). Colored voxels indicate areas where the percentage of participants with damage to a given voxel was greater in the
IMA than LCVA group (left) or low-performance than high-performance group (right). For clarity, we only plot voxels where the percentage
difference was between 40% and 60% (red), 60% and 80% (orange), and 80% to 100% (yellow).
Jax, Buxbaum, and Moll 2071
Comparison of High- and Low-performing Groups
To examine the lesions associated with performance on
the present tasks, we first examined whether lesion
volume was predictive of overall performance. It was
not (r = À.19, p = .52). Next, we classified all patients
in the IMA and LCVA groups as exhibiting high or low
production performance by using a permutation statistical
approach (Goode, 2005). To do so, we first rank
ordered all participants on their mean movement production
across all tasks.5
Next, we simulated dividing the
patients into two groups along multiple points in the
ranking (e.g., between the fourth and fifth best performing
participants, between the fifth and sixth best
performing participants, etc.), with the restriction that at
least five patients be in a group. Finally, we determined
which dividing point led to the greatest difference in
behavioral scores across groups, as measured by the
independent-sample t statistic.
The grouping resulted in a low-performing group
(n = 5) consisting of the five lowest performing IMA
patients and a high-performing group (n = 9) consisting
of the remaining five IMA3
patients and all four of the
LCVAs. The difference between mean performance in
these two groups was highly significant (.79 vs. .89,
p = .0002). Finally, we used a lesion subtraction analysis
to compare the lesion loci for patients in the high- and
low-performing groups. As shown in Figure 5 (right), the
IPL (BAs 39 and 40) most strongly differentiated the two
performance groups. There was a small region within
BA 40 in which the difference between the proportion
damage in the two groups was between 80% and
100%. In addition to the IPL, damage to smaller regions
of the premotor area (BA 6), primary sensory cortex
(BA 2), superior parietal lobe (BA 7), posterior–inferior
portion of the temporal lobe (BA 37), and anterior
occipital lobe (BA 19) also showed differences between
the two groups, although these difference were usually
weaker (40–60% difference between groups).
Results of the lesion subtraction analysis were confirmed
statistically by using a nonparametric hypothesis
test4
with the four a priori regions of interest as described
earlier. Only in the IPL was the proportion of damage
significantly more extensive in the low-performing group
than the high-performing group (.50 vs. .19, p = .049).
Damage to the other three areas, as well as all other
Brodmann’s areas, was not reliably different between the
two groups ( p ! .17).
DISCUSSION
In this study we used two central issues in the study of
motor control to frame a set of predictions about
patients with IMA. IMA patients and two comparison
groups performed grasp imitation and posture imitation
movements to object- and body-relative end positions in
the presence or absence of visual feedback. The intrinsic
coordinate control account predicted that IMA patients
would be more deficient in movements to body-relative
end positions than object-relative end positions, more
deficient with postures than grasping, more deficient on
intrinsic coordinate frame movement components than
extrinsic coordinate frame components, and show no
effects of removal of vision. The correction-overreliance
hypothesis predicted no end position, task, or movement
component effects, but predicted that IMA patients
would be disproportionately disrupted by the
removal of visual feedback.
Several results deserve emphasis. First, patients with
IMA showed a significantly greater disparity than the
other two groups between movements made to bodyrelative
and object-relative targets. Second, patients with
IMA showed a greater disparity than the other groups
between imitation of meaningless postures and grasping,
as revealed in the wrist angle score. Third, both of
these findings were observed whether or not online
correction was possible. Fourth, the IMA group showed
deficits in both intrinsic (hand configuration, wrist angle)
and extrinsic (hand orientation, location) coordinateframe
movement components. Fifth, IMA patients were
more disrupted than the other groups by the removal
of vision. Sixth, the removal of visual feedback did not
differentially affect tasks predicted to rely on intrinsic
coordinate control (posture task, body-relative end positions)
as compared to tasks predicted to rely on
extrinsic coordinate control (grasp task, object-relative
end positions). Finally, the results of the movement
matching task, with no movement production component,
showed that the deficits in movement production
could not be explained by more elementary deficits in
perception or working memory.
In summary, the pattern of results was not strongly
predicted by either account under consideration to the
exclusion of the other. IMA patients exhibit behavioral
patterns consistent with both deficient intrinsic coordinate
control and overreliance upon visual feedback.
There are at least three possible explanations for these
results. The first is that deficits in intrinsic coordinateframe
control and abnormal reliance on visual feedback
are two independent deficits that sum to produce the
observed behavioral pattern in IMA.
A second, more parsimonious, explanation is that
preplanned and visual-feedback-informed aspects of
movements preferentially rely on intrinsic and extrinsic
coordinate control, respectively. For example, a grasping
movement’s hand configuration may be planned
largely in intrinsic coordinates (e.g., a time-varying sequence
of joint angles of the fingers and thumb), with
correction at the end of the movement being more
strongly specified in extrinsic coordinates (e.g., how
the distance between the thumb and fingers compares
with the size of the to-be-grasped object). On this hypothesis,
IMA patients may compensate for deficient intrinsic
coordinate control during planning with relative
2072 Journal of Cognitive Neuroscience Volume 18, Number 12
reliance on intact extrinsic coordinate control during the
correction phase.
The claim that visual-feedback-informed aspects of
movements preferentially rely on extrinsic coordinate
control is supported by data from neurologically intact
individuals. For example, Newport, Rabb, and Jackson
(2002) asked participants to match the orientation of
two handheld bars in either extrinsic coordinates (make
the bars parallel) or intrinsic coordinates (make the bars
mirror images of one another, achieved by adopting
similar configurations of the two arms). In both conditions,
vision of the arms and bars was obscured by a
barrier, and the participants’ eyes were either open or
covered with a blindfold. Even though vision provided
no task-relevant information, participants were more
accurate in the extrinsic task, but not the intrinsic task,
with eyes open than when blindfolded. These results
suggest that vision of the environment favors the representation
of spatial relationships in extrinsic coordinates.
The claim that movement planning is accomplished
primarily in intrinsic coordinates is less clearly supported.
On the one hand, data supporting this claim
were provided by Rosenbaum, Meulenbroek, and
Vaughan (1999), who asked participants to make bimanual
movements without visual feedback. When successive
movements to the same spatial location were
performed with different arms, analysis of the spatial
errors indicated that the motor system transferred final
arm configuration information (an intrinsic property),
but not final spatial location information (an extrinsic
property), across arms. This study, along with others
(Jaric, Corcos, Gottlieb, Ilic, & Latash, 1994; Jaric, Corcos,
& Latash, 1992; Martinuk & Roy, 1972), has been used to
support models of motor control that plan movement in
intrinsic coordinates (Rosenbaum et al., 2001; Feldman,
1986; Polit & Bizzi, 1978).
Other studies, however, suggest planning is at least
sometimes accomplished in extrinsic coordinates. For
example, Wang and Sainburg (2005) showed that generalization
from learning a visuomotor rotation task
occurs in extrinsic coordinates. In this study, a 308
directional rotation was introduced between a visually
presented target and the movement to reach to that
target. Generalization from training to later novel movements
involved a remapping between the target and
movement vectors (extrinsic coordinate properties) and
not a remapping between the visual target location
and the final arm posture (an intrinsic coordinate property).
Similar conclusions about extrinsic coordinate
planning have been drawn using other methodologies
(e.g., Vindras et al., 2005; Krakauer, Pine, Ghilardi, &
Ghez, 2000; Vindras & Viviani, 1998; Gordon et al., 1994;
Rosenbaum, 1980).
These conflicting results suggest the need to expand
our second explanation of the deficits in IMA to include
planning in both intrinsic and extrinsic coordinate
frames. Thus, IMA patients may have deficits in planning
in both coordinate frames, with deficits in intrinsic
planning being greater than deficits in extrinsic planning,
along with intact online correction (which may
occur primarily in extrinsic coordinates; Newport et al.,
2002). This account has the merit of accommodating the
persistent superiority of movements to object-relative as
compared to body-relative end positions even in conditions
that are highly reliant on planning (e.g., the
delayed blindfolded conditions), something the previous
explanation can not do. Directly testing these three
explanations will be a challenge for future research.
Although deficits in intrinsic coordinate control with
relatively intact extrinsic coordinate control (which may
be preferentially used for online correction) can account
for many of the behavioral patterns in IMA, several
caveats are in order. First, the present results are based
on a small sample of apraxic patients. Even though all of
the critical main effects and interactions were significant,
indicating that there was sufficient experimental power
despite the small sample size, replication of the results
are needed. Second, the proposal that apraxics have
deficits in intrinsic coordinate control does not exclude
additional deficits in some patients. For example, it is
unclear how this hypothesis could account for the
commonly observed difference between transitive (object
related) and intransitive (symbolic) pantomime and
imitation (Foundas et al., 1999; Haaland & Flaherty,
1984) without additional assumptions, namely, that
transitive pantomime requires either more intrinsic control
or allows for less online correction than does
intransitive pantomime. This account also does not explain
the deficits of many IMA patients in recognition of
skilled movements (Gonzales Rothi, Ochipa, & Heilman,
1991). Such recognition deficits are usually ascribed to
damage to or deficient access to stored representations
of movements. (The intact performance on the
movement matching task need not be inconsistent
with these previous studies of movement recognition
in apraxia because these stored representations are
probably longer lasting than the short-term representations
in working memory used to compare two meaningless
movements separated by a delay.) It is clear that
IMA is a complex syndrome, and it is unlikely that the
entire behavioral profile of most patients can be explained
without invoking several factors.
Finally, we reported two lesion analyses showing that
damage to the IPL was associated with both IMA as well
as poor performance on the present tasks. This result
is consistent with previous studies of IMA (Buxbaum,
Johnson-Frey, et al., 2005; Buxbaum, 2001; Haaland
et al., 2000; Heilman et al., 1982) as well as a recent theory
of motor control that postulates that movement planning
and online corrections are subserved by the inferior
and superior parietal lobules, respectively (Glover, 2004).
The performance-based lesion analysis also suggests
that the left IPL may be critical for intrinsic coordinate
control. Additional studies will be required to confirm or
Jax, Buxbaum, and Moll 2073
disconfirm this hypothesis. However, it is of note that
several recent studies using functional magnetic resonance
imaging (fMRI) (Parsons et al., 1995) and cognitive
neuropsychological approaches (Schwoebel & Coslett,
2005; Schwoebel, Buxbaum, & Coslett, 2004) indicate
the importance of the left IPL for body representation.
Thus, the perceptual-motor system’s representations of
the body (e.g., the ‘‘body schema’’) may plausibly be
related to intrinsic control processes.
In addition to IPL damage, we also observed significantly
more damage to the left superior temporal lobe
in the IMA group than the LCVA group (although
damage to this area was not different in the second,
performance-based lesion analysis). Even though reports
of damage to the left superior temporal lobe have not
been common in IMA, numerous neuroimaging studies
have found significant activations in this area during
movement planning (Johnson-Frey, Newman-Norlund,
& Grafton, 2005; Gerardin et al., 2000), imitation of novel
gestures (Rumiati et al., 2005; Peigneux et al., 2004), and
perception of biological motion (Decety & Grezes, 1999),
three functions that would be required to perform tasks
that are often deficient in IMA. Thus, damage to the
superior temporal lobe may contribute more to the
deficits in IMA than previously thought. However, a
second explanation for this result is that proximity of
the IPL and superior temporal lobes leads to a high
probability of co-occurring damage in both areas. Therefore,
the superior temporal lobe may not play a functional
role in IMA. Additional investigation with larger samples
of patients may shed additional light on the substrates of
action imitation (see Buxbaum, Kyle, et al., 2005).
In summary, we have presented evidence that patients
with IMA have deficits in intrinsic coordinate frame
control as well as an overreliance on visual feedback.
The precise relationship between the two deficits is
unclear and awaits further development. However, some
combination of these deficits can account for many of
the core behavioral features of IMA, including differences
between pantomime and imitation with and without
objects in hand, deficits in meaningless posture
imitation, differences between imagined and actual
grasping, and differences between grasping with and
without visual feedback. Finally, results of lesion analysis
suggest that damage to the left inferior parietal lobe
(BAs 39 and 40) plays a key role in both deficits.
Acknowledgments
This research was supported by NIDRR grant H133G030169
and NIH grants R01-NS036387 and T32-HD007425. The authors
thank Dr. H. Branch Coslett for his assistance with the lesion
analyses and for his helpful comments on an earlier draft of
this work.
Reprint requests should be sent to Laurel J. Buxbaum, Moss
Rehabilitation Research Institute, Korman Building 213, 1200 W.
Tabor Road, Philadelphia, PA 19141, or via e-mail: lbuxbaum@
einstein.edu.
Notes
1. One previous study (Hermsdorfer, Blankenfeld, & Goldenberg,
2003) compared a reaching movement to the nose
to reaching movements to objects in the environment and
found no obvious difference between the two movement
types. Critically, the authors tested only one body-directed
movement, and therefore any conclusions drawn from this
null result should be taken cautiously.
2. Although direct comparisons between the immediate and
delayed blindfolded conditions were not appropriate (because
they differed in both the presence of visual feedback and the
delay between stimulus presentation and movement initiation),
it is noteworthy that the introduction of the delay did
not change production performance relative to the immediate
initiation condition. This was confirmed using a 3 (Group: control,
LCVA, or IMA) Â 2 (Initiation: immediate or delayed) Â 2
(Movement Task: grasp or posture) Â 2 (End Position: objectrelative
or body-relative) ANOVA. The results of this ANOVA
showed no main effect of Initiation (p = .41) and no interactions
between Initiation or any of the other factors (p ! .32).
3. One patient in the apraxic group (I6) was excluded from
the lesion analysis because examination of her high-resolution
CT scan showed no localizable lesion.
4. The statistic relied on a permutation, or resampling, hypothesis
test that make no assumptions about the underlying
population distribution from which the data were sampled
(Goode, 2005). We chose to use permutation hypothesis
tests rather than other nonparametric methods (e.g., Mann–
Whitney/Wilcoxon test) because the former do not require
ranking of data, which leads to loss of information about the
magnitude differences between individuals and groups. As applied
to the analysis of two independent samples, a permutation
hypothesis test compares the observed overlap (as
measured by the t statistic) between data in two groups (of
sample size N1 and N2) to the overlap of all possible groupings
of the pooled data into groups of size N1 and N2. This is done
computationally by pooling the observed data values from
both groups and selecting samples of size N1 and N2, without
replacement, from the pooled data values. Next, the overlap of
the two sampled groups is measured using the independentsample
t statistic. This process is repeated until t values are
computed for all possible combinations of the pooled data into
samples of size N1 and N2. From this repeated sampling, a
distribution of t values is created. The p value reported describes
the probability of observing such an extreme t value for
the actual grouping of the data relative to the distribution of
t values for all possible groupings of the data.
5. In addition to the analysis comparing the lesions associated
with high- and low-performance on all tasks, we also explored
the corresponding analyses of high- and low-performance in
the tasks shown to be particularly deficient in the apraxic group
(e.g., the posture tasks, tasks with body-relative end positions,
the delayed blindfolded condition). Because grouping based
on high- and low-performance on these different tasks resulted
in nearly identical groups, we have chosen to not report these
task-specific analyses.
REFERENCES
Bullock, D., & Grossberg, S. (1988). Neural dynamics of
planned arm movements: Emergent invariants and
speed–accuracy properties during trajectory formation.
Psychological Review, 95, 49–90.
Bullock, D., Grossberg, S., & Guenther, F. (1993). A
self-organizing neural model of motor equivalent
reaching and tool use by a multi-joint arm. Journal
of Cognitive Neuroscience, 5, 408–435.
2074 Journal of Cognitive Neuroscience Volume 18, Number 12
Buxbaum, L. J. (2001). Ideomotor apraxia: A call to action.
Neurocase, 7, 445–458.
Buxbaum, L. J., Giovannetti, T., & Libon, D. (2000). The
role of the dynamic body schema in praxis: Evidence from
primary progressive apraxia. Brain and Cognition, 44,
166–191.
Buxbaum, L. J., Johnson-Frey, S. H., & Bartlett-Williams, M.
(2005). Deficient internal models for planning hand–object
interactions in apraxia. Neuropsychologia, 43, 917–929.
Buxbaum, L. J., Kyle, K. M., Grossman, M., & Coslett, H. B.
(in press). Left inferior parietal representations for skilled
hand–object interactions: Evidence from stroke and
corticobasal degeneration. Cortex.
Buxbaum, L. J., Kyle, K. M., & Menon, R. (2005). On
beyond mirror neurons: Internal representations
subserving imitation and recognition of skilled
object-related actions in humans. Brain Research, 25,
226–239.
Buxbaum, L. J., Sirigu, A., Schwartz, M. F., & Klatzky, R.
(2003). Cognitive representations of hand posture in
ideomotor apraxia. Neuropsychologia, 41, 1091–1113.
Damasio, H., & Damasio, H. R. (1989). Lesion analysis in
neuropsychology. New York: Oxford University Press.
De Renzi, E. (1985). Methods of limb apraxia examination
and their bearing on the interpretation of the disorder.
In E. A. Roy (Ed.), Neuropsychological studies of apraxia
and related disorders (pp. 45–64). Amsterdam: Elsevier.
Decety, J., & Grezes, J. (1999). Neural mechanisms subserving
the perception of human actions. Trends in Cognitive
Sciences, 3, 172–178.
Elliot, D., Helsen, W. F., & Chua, R. (2001). A century later:
Woodworth’s (1899) two component model of goal-directed
aiming. Psychological Bulletin, 127, 342–357.
Feldman, A. G. (1986). Once more on the equilibrium point
hypothesis (l model) for motor control. Journal of Motor
Behavior, 18, 17–54.
Fiez, J. A., Damasio, H., & Grabowski, T. J. (2000). Lesion
segmentation and manual warping to a reference brain:
Intra- and interobserver reliability. Human Brain Mapping,
9, 192–211.
Foundas, A. L., Macauley, B. L., Raymer, A. M., Maher, L. M.,
Rothi, L. J., & Heilman, K. M. (1999). Ideomotor apraxia
in Alzheimer disease and left hemisphere stroke: Limb
transitive and intransitive movements. Neuropsychiatry,
Neuropsychology, and Behavioural Neurology, 12,
161–166.
Gerardin, E., Sirigu, A., Lehericy, S., Poline, J., Gaymard, B.,
Marsault, Y. A., et al. (2000). Partially overlapping neural
networks for real and imagined hand movements.
Cerebral Cortex, 10, 1093–1104.
Ghafouri, M., Archambault, P. S., Adamovich, S. V., &
Feldman, A. G. (2002). Pointing movements may be
produced in different frames of reference depending
on the task demand. Brain Research, 929, 117–128.
Glover, S. (2004). Separate visual representations in the
planning and control of action. Behavioural and Brain
Sciences, 27, 3–24.
Goldenberg, G., & Hagmann, S. (1997). The meaning of
meaningless gestures: A study of visuo-imitative apraxia.
Neuropsychologia, 35, 333–341.
Goldenberg, G., Hentze, S., & Hermsdo¨rfer, J. (2004). The
effect of tactile feedback on pantomime of tool use in
apraxia. Neurology, 63, 1863–1867.
Gonzales Rothi, L. J., Ochipa, C., & Heilman, K. M. (1991).
A cognitive neuropsychological model of limb praxis.
Cognitive Neuropsychology, 8, 443–458.
Goode, P. (2005). Permutation, parametric, and bootstrap
tests of hypotheses. New York: Springer.
Gordon, J., Ghilardi, M. F., & Ghez, C. (1994). Accuracy of
planar reaching movements. I. Independence of direction
and extent variability. Experimental Brain Research, 99,
97–111.
Haaland, K. Y., & Flaherty, D. (1984). The different types
of limb apraxia errors made by patients with left vs. right
hemisphere damage. Brain and Cognition, 3, 370–384.
Haaland, K. Y., Harrington, D. L., & Knight, R. T. (1999).
Spatial deficits in ideomotor limb apraxia. A kinematic
analysis of aiming movements. Brain, 122, 1169–1182.
Haaland, K. Y., Harrington, D. L., & Knight, R. T. (2000).
Neural representations of skilled movement. Brain, 123,
2306–2313.
Heilman, K. M., & Gonzales Rothi, L. J. (1993). Apraxia.
In K. M. Heilman & E. Valenstein (Eds.), Clinical
Neuropsychology. New York: Oxford University Press.
Heilman, K. M., Gonzales Rothi, L., Mack, L., Feinberg, T.,
& Watson, R. T. (1986). Apraxia after a superior parietal
lesion. Cortex, 22, 141–150.
Heilman, K. M., Gonzales Rothi, L. J., & Valenstein, E. (1982).
Two forms of ideomotor apraxia. Neurology, 32, 342–346.
Hermsdorfer, J., Blankenfeld, H., & Goldenberg, G. (2003).
The dependence of ipsilesional aiming deficits on
task demands, lesioned hemisphere, and apraxia.
Neuropsychologia, 41, 1628–1643.
Hermsdorfer, J., Ulrich, S., Marquardt, C., Goldenberg, G.,
& Mai, N. (1999). Prehension with the ipsilesional hand
after unilateral brain damage. Cortex, 35, 139–161.
Iacoboni, M. (2005). Neural mechanisms of imitation.
Current Opinion in Neurobiology, 15, 632–637.
Ietswaart, M., Carey, D. P., Della Sala, S., & Dijkhuizen, R. S.
(2001). Memory-driven movements in limb apraxia: Is there
evidence for impaired communication between the dorsal
and the ventral streams? Neuropsychologia, 39, 950–961.
Jaric, S., Corcos, D. M., Gottlieb, G. L., Ilic, D. B., & Latash,
M. L. (1994). The effects of practice on movement
distance and final position reproduction: Implications
for the equilibrium-point control of movements.
Experimental Brain Research, 100, 353–359.
Jaric, S., Corcos, D. M., & Latash, M. L. (1992). Effects of
practice on final position reproduction. Experimental
Brain Research, 91, 129–134.
Johnson-Frey, S. H., Maloof, F. R., Newmann-Norland, R.,
Farrer, C., Inati, S., & Grafton, S. T. (2003). Actions or
hand object interactions? Human inferior frontal cortex
and action observation. Neuron, 39, 1053–1058.
Johnson-Frey, S. H., Newman-Norlund, R., & Grafton S. T.
(2005). A distributed left hemisphere network active
during planning of everyday tool use skills. Cerebral
Cortex, 15, 681–695.
Kawato, M. (1996). Bidirectional theory approach to
integration. In T. Inui & J. L. McClelland (Eds.), Attention
and performance XVI: Information integration
(pp. 335–367). Cambridge: MIT Press.
Krakauer, J. W., Pine, Z. M., Ghilardi, M. F., & Ghez, C.
(2000). Learning of visuomotor transformations for
vectorial planning of reaching trajectories. Journal of
Neuroscience, 20, 8916–8924.
Laimgruber, K., Goldenberg, G., & Hermsdorfer, J. (2005).
Manual and hemispheric asymmetries in the execution of
actual and pantomimed prehension. Neuropsychologia,
43, 682–692.
Marteniuk, R. G., & Roy, E. A. (1972). The codability of
kinesthetic location and distance information. Acta
Psychologica, 36, 471–479.
Newport, R., Rabb, B., & Jackson, S. R. (2002). Noninformative
vision improves haptic spatial perception. Current
Biology, 12, 661–1664.
Jax, Buxbaum, and Moll 2075
Parsons, L. M., Fox, P. T., Downs, J. H., Glass, T., Hirsch,
T. B., Martin, C. C., et al. (1995). Use of implicit motor
imagery for visual shape discrimination as revealed by
PET. Nature, 375, 54–58.
Peigneux, P., Van der Linden, M., Garraux, G., Laureys, S.,
Deguldre, C., Aerts, J., et al. (2004). Imaging a cognitive
model of apraxia: The neural substrate of gesturespecific
cognitive processes. Human Brain Mapping,
21, 119–142.
Poizner, H., Merians, A. S., Clark, M. A., Macauley, B., Gonzales
Rothi, L. J., & Heilman, K. M. (1998). Left hemispheric
specialization for learned, skilled, and purposeful action.
Neuropsychology, 12, 163–182.
Polit, A., & Bizzi, E. (1978). Processes controlling arm
movements in monkeys. Science, 201, 1235–1237.
Rapcsak, S. Z., Ochipa, C., Anderson, K. C., & Poizner, H.
(1995). Progressive ideomotor apraxia: Evidence for a
selective impairment in the action production system.
Brain & Cognition, 27, 213–236.
Rosenbaum, D. A. (1980). Human movement initiation:
Specification of arm, direction, and extent. Journal of
Experimental Psychology: General, 109, 444–474.
Rosenbaum, D. A., Meulenbroek, R. G., & Vaughan, J.
(1999). Remembered positions: Stored locations or
stored postures? Experimental Brain Research, 124,
503–512.
Rosenbaum, D. A., Meulenbroek, R. G. J., Vaughan, J.,
& Jansen, C. (2001). Posture-based motion planning:
Applications to grasping. Psychological Review, 10,
709–734.
Rumiati, R. I., Weiss, P. H., Tessari, A., Assmus, A., Zilles, K.,
Herzog, H., et al. (2005). Common and differential neural
mechanisms supporting imitation of meaningful and
meaningless actions. Journal of Cognitive Neuroscience,
17, 1420–1431.
Schwoebel, J., & Coslett, H. B. (2005). Evidence for multiple,
distinct representations of the human body. Journal of
Cognitive Neuroscience, 17, 543–553.
Schwoebel, J., Buxbaum, L. J., & Coslett, H. B. (2004).
Representations of the human body in the production
and imitation of complex movements. Cognitive
Neuropsychology, 21, 285–298.
Sirigu, A., Duhamel, J.-R., Cohen, L., Pillon, B., Dubois, B.,
& Agid, Y. (1996). The mental representation of hand
movements after parietal cortex damage. Science, 273,
1564–1568.
Van Thiel, E., Meulenbroek, R. G. J., & Hulstijn, W. (1998).
Path curvature in workspace and in joint space: Evidence
for coexisting coordinative rules in aiming. Motor Control, 2,
334–351.
Vindras, P., Desmurget, M., & Viviani, P. (2005). Error
parsing in visuomotor pointing reveals independent
processing of amplitude and direction. Journal of
Neurophysiology, 94, 1212–1224.
Vindras, P., & Viviani, P. (1998). Frames of reference and
control parameters in visuomanual pointing. Journal of
Experimental Psychology: Human Perception and
Performance, 24, 569–591.
Wang, J., & Sainburg, R. L. (2005). Adaptation to visuomotor
rotations remaps movement vectors, not final positions.
Journal of Neuroscience, 24, 4024–4030.
Woodworth, R. S. (1899). The accuracy of voluntary
movement. Psychological Review Monograph
Supplements, 3, 1–119.
2076 Journal of Cognitive Neuroscience Volume 18, Number 12